An Empirical Study of Detecting Effectiveness of Detecting Business Earnings Management Using Cross-Sectional Two-Stage Regressions

• Main Author: 林昱成
• Other Authors: 蕭子誼
• Format: Others
• Language: zh-TW
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/56472837604875808971

碩士 === 逢甲大學 === 會計與財稅研究所 === 90 === The main purposes of the study are to develop a new accruals model which labeled the “WHL model” and to empirically compare the relative examine specification and power effectiveness of Modified-Jones（Dechow et al.,1995）, Margin（Peasnell et al.,2000）and WHL model in detecting business income manipulations. In addition, we also develop a detecting abnormal related-party sales model-RPS model, and examine specification and power effectiveness of RPS model in detecting business related-party sales manipulations. The following conclusions are found in the results of actual certification from the research: 1.Modified-Jones, Margin, WHL and RPS models are well specified when applied to a random sample of firm-years. 2.Our findings suggest that when the expense manipulation amount was attained 7% to 8% of prior period total assets, then the WHL and Margin models are substantially more powerful than Modified-Jones model. 3.Our findings suggest that the WHL model is more powerful than Modified-Jones and Margin models in detecting business inventory manipulation. Specifically, when the inventory manipulation amount was attained 7% of prior period total assets, then the detecting power of the WHL model would be reached to 100%. 4.Regarding the models’ relative power to detecting artificially-induced sales which none cash received from accounts receivable on the end of period, it has been found that when the sales manipulation amount was attained 10% of prior period total assets, then the power of WHL and Modified-Jones models（excluding Margin model）would be reached to 100%. 5.Our findings suggest that the WHL model is more powerful than Modified-Jones and Margin models in detecting business induces sales and cash received from accounts receivable on the end of period. But all of the three models appeared low powers. 6.Our findings suggest that when related-party sales manipulation amount was attained about 20% of prior period total assets, then the detecting power of the RPS model would be reached to 100%. We can learn from the results of the actual certification that the WHL model is more powerful in detecting accruals manipulation. In additional, we suggested that the RPS model appeared well power in detecting related-party sales.